84Elky

1000+ hours and done for now. I wanted to understand how the IAC worked under AUJP ($8d), so went through the code to learn. As the project progressed, the uploaded document resulted to record the findings. Also uploaded is a companion document which is an overly commented version of JP's S_AUJP 2 source code hack in a Word .doc format. It's a work-in-process for me as I learn what the code does.

None of this may be meaningful to anyone else as some of it deals with code details; but in the end, it does at least describe how the IAC works in AUJP. Also covered:

• Fans and Warm Kickdown
• Startup Delay Period
• IAC Kickdown
• Throttle Follower

JP86SS

iTrader: (1)

Nice review , and a resounding Thank You for taking the time to figure out the details. (You even followed the stack !!)
Found more detail items than I can count so now I can't stop reading it !!!!

Did I see that correctly in that the IAC routines "still" have indexed addressing?
If so I'll need to lable them so if anyone inserts code between the indexes it won't mess up the code execution.

gp90gta

iTrader: (1)

This is fantasitc work, thank you, I've always have had a hard time understanding (not understanding) the $8d IAC logic, it just makes no sense to me, you would think that the IAC is really only needed for stall saving, TF, idle, and idle load changes, yet I could be cruising along on the highway with no heavy changes to tps or load and my IAC is move around but the car runs great. So now all I need now is a few year to study it to get a better understanging of it lol.

84Elky

What I was referencding in the comment was not a "jump". Rather it is where a varible's address is loaded to the X register via 'ldx' creating a pointer to a memory addess so that future references "index" off of it to point to another variable. For example:
Here, the variable represented by '0x1D,x' is L8645, obtained by "offsetting" from the starting index/pointer 0x8628 by 0x1D (0x8628 + 0x1D = 0x8645).

Really makes the code hard to follow. I can't in my wildest dreams understand why GM would have programmed it this way except in one spot where things change whether in gear or P/N. Smells like a contract programmer working by the hour to make things complicated. But that's why I added all the comments where these indexes were used. Would obviously have been better if the actual variable name was used, but it is what it is.

Here are all the places I found where a variable was used as an index pointer:
L8628 - 3 times (LC2A5, LC3DF+6, LE61E)
L862C - 1 time (LC439+2)